Final Report Summary - DIASPORA (Drift In Amorphous Semiconductors - A Partnership Of Rüschlikon and Aachen)
Project Website: http://www.iapp-diaspora.eu/index.html
Phase change memory (PCM) is arguably the most advanced emerging non-volatile memory technology. It is also finding applications in non-von Neumann computing. It is enabled by the large resistance contrast between the amorphous and crystalline states in phase change materials. The amorphous phase offers high electrical resistivity, while the crystalline phase exhibits resistivity values that can be three or four orders of magnitude lower. To “SET” the memory cell into its low-resistance state, an applied electrical pulse heats a large portion of the cell above the crystallization temperature of the material; to “RESET” it, a larger electrical current is applied, melting the central portion of the cell. If the RESET pulse is cut off abruptly enough, the molten material quenches into the amorphous phase, producing a cell in the high-resistance state. Read operations are performed by measuring the device resistance at low voltage so that the device state is not perturbed. The large resistance contrast between the amorphous and crystalline states enables storage of more than 1 bit of data per cell, using 2N analog resistance states to store N bits per cell. This allows an increase in effective density, much like MLC (Multi-Level Cell) Flash, without decreasing the feature size. However, there is an unwanted property of PCM that complicates the implementation of MLC namely the resistance “drift” after programming. The device resistance increases after programming due to structural relaxation of the amorphous phase of the material.
The goal of DIASPORA was to gain a deep understanding of the underlying physics of this resistance drift. We aimed at developing a microscopic picture of the temporal evolution of electrical transport in melt-quenched amorphous phase change materials. For this, we had to connect the atomic structure and structural relaxation with electrical transport and resistance drift through the density of states (DoS). The research methodology was as follows:
1) Perform a detailed study of the low and high field electrical transport and its temporal evolution
2) By employing various spectroscopic techniques gain insights into the density of states and make links between the DoS and the electrical transport.
3) Establish the link between atomic structure and defects to the DoS
4) Finally link structural relaxation with the temporal evolution of electrical transport
5) Arrive at new materials for phase-change memory applications
The electrical transport properties (e.g. the Seebeck effect (WP1&WP2) or the conductivity at high electric fields (WP3)) can help to characterize the important aspects of on-state dynamics and resistance drift. We employed spectroscopy methods to gain insight into the electronic density of states (WP4 & WP5). We aimed to perform MD simulations to study the influence of structural properties on resistance drift (WP6) and to arrive at new phase-change materials.
During the course of the project, we made progress on all the work packages.
The research highlights are listed below:
A comprehensive thermoelectric model was developed to capture the characteristics of phase change memory devices. A remarkable achievement was the satisfying matching of simulation and experimental data for doped-Ge2Sb2Te5 memory cells, validating the accuracy of the proposed model. Another key insight was the significant role played by the various thermo-electric components in addition to Joule heating. The reason being the large temperature gradients created in the nanoscale devices. The study was presented as a talk at the SISPAD conference in Washington DC, USA in September 2015.
With regards to electrical transport, we thoroughly investigated the low field regime of the current-voltage characteristics in amorphous phase change materials. We developed a unified model based on multiple-trapping transport together with 3D Poole–Frenkel emission from a two-center Coulomb potential to describe the so-called sub-threshold transport regime. This result was published in the New Journal of Physics. We also discovered a new conduction regime at higher fields and this study was reported in the Journal of Applied Physics. The temporal evolution of electrical transport on a micro-second time scale was also studied employing nanoscale line-cells fabricated at RWTH. The results were presented at E\PCOS 2014 in Marseille.
To gain insights into the nature of the DoS for the various phase change materials, dark- and photoconductivity measurements were conducted over a wide range of temperatures and light flux intensities. These measurements together with simulations provided strong indications for the DoS of
AIST being substantially different from the other phase change materials. These results were presented at the MRS Spring Meeting 2015 in San Francisco. To study the impact of structural relaxation on the DoS and in particular the bandgap, infrared spectra were measured on amorphous thin films of the three different phase change materials (GeTe, GST, AIST). We found a widening of the bandgap upon annealing accompanied by a decrease of the optical dielectric constant epsilon infinity for all three materials. Quantitative comparison with existing experimental data from the RWTH work group revealed that the temporal evolution of bandgap and activation energy for electrical conductivity can be decoupled. This phenomenon regarding the link between the DoS and electrical transport demonstrates a possibility to identify new phase change materials with reduced resistance drift. The study was published in Nature Scientific Reports and was presented as oral contribution to the E\PCOS 2015 conference in September in Amsterdam.
High quality modulated photo-conductivity (MPC) measurements on microscale GeSbTe and AgInSbTe devices were performed, including an advanced measurement methodology with varying light fluxes. The in-depth analysis of the measurement results for the Gaussian defect in GeSbTe compare well to results found in literature. However, while previous studies consider the capture coefficients of the valence bandtail states in amorphous PCM as constant, our results clearly indicate that these states in GeSbTe and AgInSbTe exhibit non-constant capture characteristics. This result is a striking input for future efforts to understand the link between the density of states and electrical conduction, since the latter one is naturally governed by the trap and release processes of charge carriers.
Ab Initio Molecular Dynamics (AIMD) simulations were performed to study the effect of rapidly quenching molten phase-change materials to lower temperatures. The observation of the following crystallization process under varying quenching conditions resulted in the central outcome the research: the dependence of the stability of an amorphous state on holding temperature, quenching rate and density or stress respectively. Those simulations were accompanied by a thorough series of melt-quenching experiments on according nano-scale devices, which show the potential of a class of materials that has so far not been seriously considered for applications. These results appeared on the cover of the prestigious journal Nature Materials (Aug. 2018). Over the four years, we have made significant steps towards a deep understanding of electrical transport and resistance drift in amorphous phase change materials. Our work is expected to have a significant impact on MLC PCM technology and potentially even other emerging applications of phase change memory devices such as non-von Neumann computing.
Phase change memory (PCM) is arguably the most advanced emerging non-volatile memory technology. It is also finding applications in non-von Neumann computing. It is enabled by the large resistance contrast between the amorphous and crystalline states in phase change materials. The amorphous phase offers high electrical resistivity, while the crystalline phase exhibits resistivity values that can be three or four orders of magnitude lower. To “SET” the memory cell into its low-resistance state, an applied electrical pulse heats a large portion of the cell above the crystallization temperature of the material; to “RESET” it, a larger electrical current is applied, melting the central portion of the cell. If the RESET pulse is cut off abruptly enough, the molten material quenches into the amorphous phase, producing a cell in the high-resistance state. Read operations are performed by measuring the device resistance at low voltage so that the device state is not perturbed. The large resistance contrast between the amorphous and crystalline states enables storage of more than 1 bit of data per cell, using 2N analog resistance states to store N bits per cell. This allows an increase in effective density, much like MLC (Multi-Level Cell) Flash, without decreasing the feature size. However, there is an unwanted property of PCM that complicates the implementation of MLC namely the resistance “drift” after programming. The device resistance increases after programming due to structural relaxation of the amorphous phase of the material.
The goal of DIASPORA was to gain a deep understanding of the underlying physics of this resistance drift. We aimed at developing a microscopic picture of the temporal evolution of electrical transport in melt-quenched amorphous phase change materials. For this, we had to connect the atomic structure and structural relaxation with electrical transport and resistance drift through the density of states (DoS). The research methodology was as follows:
1) Perform a detailed study of the low and high field electrical transport and its temporal evolution
2) By employing various spectroscopic techniques gain insights into the density of states and make links between the DoS and the electrical transport.
3) Establish the link between atomic structure and defects to the DoS
4) Finally link structural relaxation with the temporal evolution of electrical transport
5) Arrive at new materials for phase-change memory applications
The electrical transport properties (e.g. the Seebeck effect (WP1&WP2) or the conductivity at high electric fields (WP3)) can help to characterize the important aspects of on-state dynamics and resistance drift. We employed spectroscopy methods to gain insight into the electronic density of states (WP4 & WP5). We aimed to perform MD simulations to study the influence of structural properties on resistance drift (WP6) and to arrive at new phase-change materials.
During the course of the project, we made progress on all the work packages.
The research highlights are listed below:
A comprehensive thermoelectric model was developed to capture the characteristics of phase change memory devices. A remarkable achievement was the satisfying matching of simulation and experimental data for doped-Ge2Sb2Te5 memory cells, validating the accuracy of the proposed model. Another key insight was the significant role played by the various thermo-electric components in addition to Joule heating. The reason being the large temperature gradients created in the nanoscale devices. The study was presented as a talk at the SISPAD conference in Washington DC, USA in September 2015.
With regards to electrical transport, we thoroughly investigated the low field regime of the current-voltage characteristics in amorphous phase change materials. We developed a unified model based on multiple-trapping transport together with 3D Poole–Frenkel emission from a two-center Coulomb potential to describe the so-called sub-threshold transport regime. This result was published in the New Journal of Physics. We also discovered a new conduction regime at higher fields and this study was reported in the Journal of Applied Physics. The temporal evolution of electrical transport on a micro-second time scale was also studied employing nanoscale line-cells fabricated at RWTH. The results were presented at E\PCOS 2014 in Marseille.
To gain insights into the nature of the DoS for the various phase change materials, dark- and photoconductivity measurements were conducted over a wide range of temperatures and light flux intensities. These measurements together with simulations provided strong indications for the DoS of
AIST being substantially different from the other phase change materials. These results were presented at the MRS Spring Meeting 2015 in San Francisco. To study the impact of structural relaxation on the DoS and in particular the bandgap, infrared spectra were measured on amorphous thin films of the three different phase change materials (GeTe, GST, AIST). We found a widening of the bandgap upon annealing accompanied by a decrease of the optical dielectric constant epsilon infinity for all three materials. Quantitative comparison with existing experimental data from the RWTH work group revealed that the temporal evolution of bandgap and activation energy for electrical conductivity can be decoupled. This phenomenon regarding the link between the DoS and electrical transport demonstrates a possibility to identify new phase change materials with reduced resistance drift. The study was published in Nature Scientific Reports and was presented as oral contribution to the E\PCOS 2015 conference in September in Amsterdam.
High quality modulated photo-conductivity (MPC) measurements on microscale GeSbTe and AgInSbTe devices were performed, including an advanced measurement methodology with varying light fluxes. The in-depth analysis of the measurement results for the Gaussian defect in GeSbTe compare well to results found in literature. However, while previous studies consider the capture coefficients of the valence bandtail states in amorphous PCM as constant, our results clearly indicate that these states in GeSbTe and AgInSbTe exhibit non-constant capture characteristics. This result is a striking input for future efforts to understand the link between the density of states and electrical conduction, since the latter one is naturally governed by the trap and release processes of charge carriers.
Ab Initio Molecular Dynamics (AIMD) simulations were performed to study the effect of rapidly quenching molten phase-change materials to lower temperatures. The observation of the following crystallization process under varying quenching conditions resulted in the central outcome the research: the dependence of the stability of an amorphous state on holding temperature, quenching rate and density or stress respectively. Those simulations were accompanied by a thorough series of melt-quenching experiments on according nano-scale devices, which show the potential of a class of materials that has so far not been seriously considered for applications. These results appeared on the cover of the prestigious journal Nature Materials (Aug. 2018). Over the four years, we have made significant steps towards a deep understanding of electrical transport and resistance drift in amorphous phase change materials. Our work is expected to have a significant impact on MLC PCM technology and potentially even other emerging applications of phase change memory devices such as non-von Neumann computing.